pharmaceuticals-exploration-zhcn

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt instructs embedding an API key directly into a command/connector URL (shows an sk-... example and asks users to place "你的API_KEY" into the claude mcp add URL), which requires secrets to be copied verbatim into commands and thus risks exposure/exfiltration.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). 在运行时该技能会根据用户输入先识别药物/靶点/疾病并调用 Patsnap MCP（如 ls_target_fetch、ls_drug_search、ls_patent_search、ls_clinical_trial_search 等）去读取对应数据库条目详情，从而让用户提供的任意 free text 影响被检索到的外部（Patsnap）内容并由 LLM 读取。